What library are you using for CUDA matrix computing? Or is there? Everyone seems to write this for himself.
For regular processors, I use Eigen . What about GPUs?
For dense matrix operations, you can consider CUBLAS (supplied with the CUDA Toolkit), Magma, and CULAtools .
For sparse matrix operations, consider CUSPARSE (supplied with the CUDA Toolkit) and CUSP .
?CUSP - .Magma .